基于CVM的入侵检测  

Intrusion Detection System using Core Vector Machine

作　　者：庞雄昌[1] 王喆[1] 韩鲲[1] 

机构地区：[1]西安通信学院一系,陕西西安710106

出　　处：《微计算机信息》2008年第18期45-46,24,共3页Control & Automation

摘　　要：本文提出了基于CVM算法的入侵检测方法,这种方法先对样本集求其MEB问题,MEB问题的解就是决定分类超平面的支持矢量,然后再根据支持矢量的分布对网络的入侵行为进行分类。通过用KDD99数据的验证,证明了这种方法的有效性和可行性。Intrusion Detection System based on core vector machine is presented in this paper. Firstly ,the minimum enclosing ball of a training data set is solved by MEB （Minimum Enclosing Ball） algorithm, then the optimal separating hyperplane is constructed by the solutions of the core sets. According to the distribution of the core sets, the different intrusion actions can be detected. The related experiment indicates that the algorithm is feasible and effective.

关 键 词：核心矢量机 核心集 入侵检测 

分 类 号：TP393.08[自动化与计算机技术—计算机应用技术]